Guild Launch News
UPDATE: "Lost Connection to Server" = Bug in MySQL
We're currently researching an issue that cropped up last night where when executing the Rapid Raid members query people were intermittently receiving an error reporting "Lost Connection to Server While Executing Query". I made some changes to the database server last night to see if they would fix the issue and it did seem to help. However, a couple of other users reported the same issue after the change was made indicating that it wasn't a complete fix. Further research overnight indicates that the issue is a bug in mySQL.
The server is executing that very large query and leaking memory as it executes it, then under load when many people are executing that query (like a Friday when it is an active raid night) the server runs out of memory and restarts the database process (just the process not the server, so it only takes a second or 2 to come back up).
This is a bug in mySQL that was apparently fixed in a recent version. However, we can't upgrade the database version without considerable regression testing so we are working this weekend to rewrite the members query instead to work around the limit that mySQL is reaching.
We'll have something out soon.
-Stephen
The server is executing that very large query and leaking memory as it executes it, then under load when many people are executing that query (like a Friday when it is an active raid night) the server runs out of memory and restarts the database process (just the process not the server, so it only takes a second or 2 to come back up).
This is a bug in mySQL that was apparently fixed in a recent version. However, we can't upgrade the database version without considerable regression testing so we are working this weekend to rewrite the members query instead to work around the limit that mySQL is reaching.
We'll have something out soon.
-Stephen
UPDATE: 12/08/2007 11:20 am EST
We're continuing to investigate the database instability. I have engaged some additional database people at our datacenter to look into it. I will keep you all updated.
-Stephen
We're continuing to investigate the database instability. I have engaged some additional database people at our datacenter to look into it. I will keep you all updated.
-Stephen
UPDATE: 12/8/2007 12:34
We've adjusted some memory settings and a rapid raid query and are monitoring the situation to see if it has resolved things.
-Stephen
We've adjusted some memory settings and a rapid raid query and are monitoring the situation to see if it has resolved things.
-Stephen